Research map / Architectures and efficiency / RNNs and normalization
Morphological Segmentation with Window LSTM Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace character-level LSTM with window-based LSTM (W-LSTM) | LSTM · Hebrew S&B 100% | F1: 62.9 -> 91.9 | Helped |
| replace character-level LSTM with window-based LSTM (W-LSTM) | LSTM · Arabic S&B 100% | F1: 73.3 -> 93.6 | Helped |
| process whole word jointly (MW-LSTM) instead of single window | W-LSTM · Arabic S&B 100% | F1: 93.6 -> 96.6 | Helped |
| process whole word jointly (MW-LSTM) instead of single window | W-LSTM · Hebrew S&B 100% | F1: 91.9 -> 92.0 | Helped |
| add backward pass (bidirectional) to MW-LSTM | MW-LSTM · Hebrew S&B 100% | F1: 92.0 -> 92.2 | No clear effect |
| add backward pass (bidirectional) to MW-LSTM | MW-LSTM · Arabic S&B 100% | F1: 96.6 -> 95.5 | Hurt |
| add backward pass (bidirectional) to MW-LSTM | MW-LSTM · Hebrew S&B 25% | F1: 88.2 -> 93.2 | Helped |
| ensemble of 10 BMW-LSTM models | BMW-LSTM · Hebrew S&B 100% | F1: 92.2 -> 95.2 | Helped |
| BMW-LSTM (no feature engineering) vs feature-rich CRF | CRF · Hebrew S&B 25% | F1: 90.6 -> 93.2 | Helped |
| window LSTM models vs feature-rich CRF | CRF · Arabic S&B 100% | F1: 97.8 -> 96.6 | Hurt |
Closest papers by meaning
- A Morphology-Aware Network for Morphological Disambiguation (AAAI 2016 · similarity 0.87)
- A Latent Morphology Model for Open-Vocabulary Neural Machine Translation (ICLR 2020 · similarity 0.87)
- Character-Aware Neural Language Models (AAAI 2016 · similarity 0.85)
- Switch-LSTMs for Multi-Criteria Chinese Word Segmentation (AAAI 2019 · similarity 0.84)
- Fairness in Representation for Multilingual NLP: Insights from Controlled Experiments on Conditional Language Modeling (ICLR 2022 · similarity 0.83)
- Vietnamese Words Are Not Constructed from Syllables: Rethinking the Role of Word Segmentation in Natural Language Processing for Vietnamese Texts (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.